Self-dispersed crumpled graphene balls in oil forfriction and wear reductionXuan Doua, Andrew R. Koltonowa, Xingliang Heb, Hee Dong Jangc, Qian Wangb,1, Yip-Wah Chunga,b,1,and Jiaxing Huanga,1

aDepartment of Materials Science and Engineering, Northwestern University, Evanston, IL 60208; bDepartment of Mechanical Engineering, NorthwesternUniversity, Evanston, IL 60208; and cRare Metals Research Center, Korea Institute of Geoscience & Mineral Resources, Daejeon, 305-350, South Korea

Edited by Alexis T. Bell, University of California, Berkeley, CA, and approved December 24, 2015 (received for review October 23, 2015)

Ultrafine particles are often used as lubricant additives becausethey are capable of entering tribological contacts to reduce frictionand protect surfaces from wear. They tend to be more stable thanmolecular additives under high thermal and mechanical stressesduring rubbing. It is highly desirable for these particles to remainwell dispersed in oil without relying on molecular ligands. Borrowingfrom the analogy that pieces of paper that are crumpled do notreadily stick to each other (unlike flat sheets), we expect that ultrafineparticles resembling miniaturized crumpled paper balls should self-disperse in oil and could act like nanoscale ball bearings to reducefriction and wear. Here we report the use of crumpled graphene ballsas a high-performance additive that can significantly improve thelubrication properties of polyalphaolefin base oil. The tribologicalperformance of crumpled graphene balls is only weakly dependenton their concentration in oil and readily exceeds that of other carbonadditives such as graphite, reduced graphene oxide, and carbon black.Notably, polyalphaolefin base oil with only 0.01–0.1 wt % of crum-pled graphene balls outperforms a fully formulated commercial lubri-cant in terms of friction and wear reduction.

lubrication | tribology | aggregation-resistant particles | graphene

Lubricants reduce friction between contacting surfaces andthus increase the energy efficiency of engines and other

machines. They can also reduce wear, thereby extending the lifeof tribological components. Many types of ultrafine particles,here loosely defined as submicrometer-sized particles, have beenstudied as lubricant additives (1, 2) because they can entercontact regions between sliding surfaces and protect them fromdirect mechanical contact (2–4). This makes ultrafine particleseffective for reducing friction and wear in the boundary regime,such as during startup or low-speed operation of an engine, whenthe lubricant film at these contacts is too thin to prevent directmetal–metal contact (2–4). Further, because of the high shearstresses and sometimes high local temperatures at these contacts,molecular additives can rub off, decompose, or simply fail toprovide a sufficiently thick coverage for friction reduction andwear protection (2–4). Therefore, ultrafine particles are appealingby virtue of their size and their chemical and thermal stability undertribological conditions. However, it is challenging to disperse ul-trafine particles in lubricating oils. Typically, this requires surfacefunctionalization with surfactant-like substances, which themselvesare prone to degradation under tribological conditions, leading tounstable lubrication properties (2). Therefore, it would be highlydesirable if ultrafine particle additives could remain self-dispersed inlubricant oil without the use of ligands.This work was inspired by the analogy with crumpled paper. A

crumpled piece of paper in the shape of a ball has a rough sur-face texture, which reduces the area of contact when placed ontop of a flat sheet of paper or in an assembly of crumpled paperballs, resulting in minimal adhesion. Because of the multiplefolding, crumpled paper balls become strain-hardened (and thusstiffer) under mechanical stress, so they can largely maintain theirshape and shape-induced nonstick properties (5–7). One mightexpect, then, that ultrafine particles in the shape of crumpled paper

balls could be well dispersed in oil and have superior lubricationproperties. Such miniaturized crumpled structures were first re-alized with graphene-based materials using an aerosol capillarycompression approach (6). Just as how a crumpled paper ball ismade by isotropically compressing a sheet of paper with one’shands, flat graphene-based sheets suspended in nebulized aerosoldroplets are isotropically compressed during solvent evaporation,leading to the final crumpled ball morphology. The resultant sub-micrometer-sized crumpled graphene balls indeed have propertiesanalogous to those of crumpled paper, including strain hardeningand aggregation resistance. The morphology of crumpled grapheneballs is highly stable in the solid state and when dispersed in liquids.They do not unfold or collapse even after heating or pelletizing.Because they consistently do not form intimate contact with eachother, their interparticle van der Waals attraction is so weak thatthey can be individually dispersed in nearly any solvent, includinglubricant oils, without the need for any chemical functionalization(8). Despite their compact appearance, crumpled graphene ballshave a great deal of free volume and solvent-accessible surface areainside, making them an effective absorber of oil, which could bereleased upon compression, ensuring uninterrupted wetting of thecontact area. These properties should make them highly desirablefor tribological applications. In this work, we demonstrate thatcrumpled graphene ball is indeed a superior friction modifiercompared with other carbon additives such as graphite powders,chemically exfoliated graphene sheets, and carbon black (9–16).Remarkably, base oil with just 0.01–0.1 wt % of crumpled grapheneballs is more effective in friction and wear reduction than a fullyformulated commercial lubricant.

Significance

Aggregation is a major problem for ultrafine particle additivesin lubricant oil because it reduces the effective particle con-centrations, prevents particles from entering the contact areaof working surfaces, and leads to unstable tribological per-formance. Molecular ligands can help the particles to disperse,but they tend to degrade under the harsh tribological condi-tions. Therefore, self-dispersed particles without the need forsurfactant are highly desirable. We report here, for the firsttime to our knowledge, such type of ultrafine particles made ofcrumpled, paper-ball–like graphene, which indeed can self-dispersein lubricant oil, and exhibit stable and superior tribologicalperformances.

Author contributions: X.D., Q.W., Y.-W.C., and J.H. designed research; X.D. and X.H.performed research; X.D. and H.D.J. contributed new reagents/analytic tools; X.D.,A.R.K., X.H., Q.W., Y.-W.C., and J.H. analyzed data; and X.D., A.R.K., Q.W., Y.-W.C., andJ.H. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence may be addressed. Email: qwang@northwestern.edu,ywchung@northwestern.edu, or jiaxing-huang@northwestern.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1520994113/-/DCSupplemental.

1528–1533 | PNAS | February 9, 2016 | vol. 113 | no. 6 www.pnas.org/cgi/doi/10.1073/pnas.1520994113

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

Results and DiscussionDispersion and Aggregation-Resistant Properties of Crumpled GrapheneBalls. The tribological performance of crumpled graphene balls wasinvestigated in comparison with three other widely studied carbonadditives: graphite platelets, reduced graphene oxide sheets (r-GO,also known as chemically modified graphene), and carbon black.Powders of these carbon materials (0.01–0.1 wt %) were sonicated inthe lubricant base oil (polyalphaolefins type 4, PAO4) until theywere fully dispersed with no residual solids remaining. All fouradditives can initially disperse in the base oil right after sonication(Fig. 1A). However, agglomeration was apparent in the dispersionsof graphite platelets, r-GO sheets, and carbon black powders aftera few hours. After 20 h, crumpled graphene balls were still dis-persed in the oil, but the other three carbon materials were fullysedimented (Fig. 1B). The microstructures of the four carbon ad-ditives were observed with the scanning electron microscope(SEM). The sonicated graphite platelets are typically around 1–3μm in lateral dimension and 40–60 nm in thickness (Fig. 1C). Al-though they disperse initially, they are prone to aggregation due totheir flat, disk-like shape, which can form intimate interparticlecontact and generate strong attraction. Similarly, the r-GO sheetsalso tend to restack to form large chunks a few hours after soni-cation (Fig. 1D). The primary particles in carbon black powders areabout 50 nm in diameter. They aggregate into micrometer-sizedclusters, which can be broken down to submicrometer pieces bysonication (Fig. 1E). Carbon black powders can stay dispersed forabout 5–10 h. The dispersion of crumpled graphene balls, whichare around 500 nm in diameter, is most stable, because their shapeprevents them from forming tight stacking, hence preventing ag-gregation (Fig. 1F). After a few days, all of the dispersed carbonadditives shown in Fig. 1 A and B would sediment in the oil. Uponshaking, sedimented graphite platelets, r-GO sheets, and carbonblack powders can temporarily suspend in oil. Optical microscopyobservation (Fig. S1) reveals that these three suspensions containlarge, persistent micrometer-sized aggregates. Therefore, they willprecipitate again quickly. In contrast, crumpled graphene balls canremain finely dispersed.A pin-on-disk tribometer (Fig. 2 A and B) was used to study

the tribological properties of the carbon-based additives inPAO4 base oils (16, 17). The pin and the disk were made of M50steel ball (Ø 9.53 mm, surface roughness Ra ∼17 nm) andE52100 steel (Ø 30 cm, Ra ∼5 nm), respectively. To keep thetribological test in the boundary lubrication regime, testing pa-rameters were chosen to ensure that the thickness of the lubri-cant film was smaller than the surface roughness (18). According

to the Hamrock–Dowson equation(19), this condition can bemet by applying a 10-N load to the pin while the linear slidingspeed of the disk relative to the pin is 10 mm/s. The maximumHertzian contact pressure was about 1 GPa (20, 21). To test ifthe crumpled graphene balls can sustain this high pressure, staticcompression experiments were performed first by using the samepin-on-disk configuration with a 10-N load. As shown in Fig.S2A, crumpled graphene balls drop-casted onto a polished steeldisk formed a uniform film. The steel ball left a nearly circularcontact area of around 300 μm in diameter. The SEM overviewimage (Fig. S2B) of this area shows that some patches of crumpledgraphene balls were removed with the ball, exposing the surface ofthe steel disk. However, crumpled graphene balls remaining in thecontact area did not appear flattened or severely deformed (Fig. S2C–E). The resistance of crumpled graphene ball to compression isattributed to its strain-hardening property due to the multiple foldscreated within the ball during its formation. Upon further com-pression, more folds can be generated, leading to increased stiffness.Results shown in Fig. S2 suggest that crumpled graphene balls cansurvive the high pressure while remaining their crumpled ball shape.

Self-Dispersed Crumpled Graphene Balls as Friction Modifiers. Frictiontesting results are shown in Fig. 2 C–F. Steel surfaces lubricated withthe base oil display a typical type of run-in behavior. Initially, thefriction coefficient increases due to the wear of asperities at thecontact surfaces. Eventually, development of wear tracks increasesthe contact area and reduces the contact pressure, hence loweringthe friction coefficient. The friction curve for r-GO is similar to thatof the base oil. Among all of the carbon-based additives, crumpledgraphene ball gives the lowest friction coefficient: only 0.01 wt % ofcrumpled graphene is needed to reduce the friction coefficient by20% compared with the base oil. In practice, a good additive shouldmaintain consistent performance over a range of concentrations sothat local concentration fluctuations and/or material loss do notdisrupt the functionality of the additive. Therefore, tests were alsoconducted at higher loading, 0.1 wt %. This high concentration (Fig.2D) results in no significant improvement in friction performancefor r-GO, whereas graphite and carbon black display poorer per-formance, probably due to easier aggregation at higher concentra-tion. Such larger aggregates are likely to be poorly dispersed andcannot protect contacting surfaces. In fact, interaction among theselarger aggregates could induce jamming during the friction test,leading to increased friction coefficient (22). By contrast, crumpledgraphene displays consistent friction behavior at this higher con-centration. Its self-dispersion and aggregation-resistant properties

Fig. 1. Dispersion properties of four carbon additives in the lubricating oil. Dispersion of four carbon additives, all at 0.1 wt %, in PAO4 (A) immediately aftersonication and (B) 20 h after sonication; (C–F) SEM images of drop-casted powders of graphite, r-GO, carbon black, and crumpled graphene balls, respectively.The crumpled graphene balls stay dispersed due to their aggregation-resistant properties. The solid content for all of the dispersions is 0.1 wt %.

Dou et al. PNAS | February 9, 2016 | vol. 113 | no. 6 | 1529

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

are primarily responsible for this consistently low friction over thisrange of concentration.After the friction tests, the wear surfaces were imaged by

SEM. Some carbon-based particles were left on the wear track.Severe aggregation was observed for all carbon-based additives(Fig. 3 A–C) except crumpled graphene balls (Fig. 3D). Intactcrumpled graphene balls can be seen filling the grooves of thewear tracks and are separated from each other clearly. Evi-dently, the strain-hardening property of crumpled nano-structure (6) prevents excessive deformation and damageduring friction.

Wear Reduction by Self-Dispersed Crumpled Graphene Balls. In additionto the substantial friction reduction, noteworthy improvements inwear reduction were also observed in our experiments. In Fig. 3,SEM images of the wear tracks already suggest that oil modi-fied with crumpled graphene balls is qualitatively more effectivein reducing wear than the other three carbon additives. Quan-titative examination of the wear tracks was conducted with whitelight interferometer, which generated a 3D map of the surface.Wear coefficients for oils modified with additives at two differentconcentrations are shown in Fig. 4 A and B. For both 0.01 wt %and 0.1 wt %, crumpled graphene ball is significantly more ef-fective than other carbon additives for wear reduction. The ap-parent wear coefficient difference in pure base oil between thetwo tests arises from the different testing times––because the mostsignificant wear tends to occur at the beginning of the test, the wearcoefficient calculated from the longer test should be lower. r-GO

sheets, which did not reduce the coefficient of friction greatly(Fig. 2), also failed to provide effective wear reduction; it showed thesame apparent decrease in wear coefficient as the base oil whentested at a higher concentration for a longer time. Similar to whatwas observed for friction coefficient, the wear reduction performance

Fig. 3. SEM images of remaining carbon additives in the wear tracks aftertribological tests. (A) Graphite, (B) r-GO, (C) carbon black, and (D) crumpledgraphene balls. Only crumpled graphene balls remain aggregation-free.(Scale bar in the Inset, 200 nm.)

Fig. 2. Coefficient of friction using PAO4 base oil with and without carbon additives. (A) Schematic illustration and (B) photo showing the pin-on-diskgeometry of the tribometer; C and D show the variation of coefficient of friction as a function of time using PAO4 base oil and with 0.01 wt % and 0.1 wt %carbon-based additives, respectively. The corresponding bar charts in E and F show the friction values averaged over the entire duration of the test. For bothconcentrations, crumpled graphene balls are found to be the most effective carbon additive for friction reduction.

1530 | www.pnas.org/cgi/doi/10.1073/pnas.1520994113 Dou et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

of graphite platelets and carbon black additives also degrades athigher concentrations. Meanwhile, varying the concentration hascomparatively little effect on the wear coefficient for the lubricantwith crumpled graphene balls. A detailed analysis of the wear tracksin Fig. 4D is presented in Fig. S3. Compared with the other threeadditives, crumpled graphene balls generated shallower (0–0.5 μm),narrower (0–4 μm), and more uniform wear tracks. In contrast, weartracks generated by other carbon additives are deeper (1–3.5 μm)and wider (12–20 μm), with a broader size distribution. It is signifi-cant that the use of crumpled graphene ball prevents the formationof wear tracks larger or deeper than 10 μm, because such wear trackstend to generate large debris that can inflict severe abrasive wear(23–25). Compared with the use of the base oil, the self-dispersedcrumpled graphene additives are able to eliminate wear by ∼85%(red bars in Fig. 4).

Benchmarking Against Fully Formulated Commercial Lubricant. Thebase oil modified with 0.1 wt % crumpled graphene balls wasalso tested for comparison with a PAO-based commercial lu-bricant 5W30 (Fig. 5A). Both 5W30 and crumpled-graphene-ball–modified PAO4 outperform the base oil, with comparable

coefficients of friction. 5W30 has organic molecular frictionmodifiers which bind to the metal surface and decrease adhe-sion, making it effective for friction reduction. However, crum-pled graphene balls are more effective in wear reduction. Thedifference is revealed by the wear track profiles shown in Fig. 5 Cand D. The surface lubricated by 5W30 still yielded deep and widewear tracks at tens of μm scale. However, crumpled graphene ballscan provide better protection of the surfaces, leaving a muchsmoother wear track, as shown by the profile (Fig. 5D).

ConclusionsIn summary, addition of crumpled graphene balls to PAO4 baseoil results in superior friction and wear performance, due largelyto their aggregation-resistant property. This unique propertymakes them more stable in the base oil than other carbon-basedmaterials, such as graphite, r-GO, and carbon black. Aggregationmakes other carbon-based materials lose their ability to preventthe contact of two surfaces, negatively impacting their tribologicalperformance. In contrast to other carbon additives, whose tri-bological properties are sensitive to their concentration, crumpledgraphene balls deliver consistently good performance between

Fig. 4. Wear coefficients of PAO4 base oil with and without carbon additives. The bar charts in A and B compare the wear coefficients of the base oil itselfand samples with 0.01 wt % and 0.1 wt % carbon additives, respectively. The corresponding 3D profile images of the wear tracks are shown in C and D. It isevident that crumpled graphene balls can better protect the steel surface from wear.

Dou et al. PNAS | February 9, 2016 | vol. 113 | no. 6 | 1531

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

0.01 wt % and 0.1 wt % concentration. It was found that crumpledgraphene balls reduce friction coefficient and wear coefficient byabout 20% and 85%, respectively, with respect to the base oil.Furthermore, base oil modified with crumpled graphene ballsalone outperforms a fully formulated 5W30 lubricant in terms offriction and wear reduction. The combination of aggregationresistance, self-dispersion, and mechanical properties of crumpledgraphene particles makes them an attractive material for tribologicalapplications.

Materials and MethodsPreparation of Tested Materials. Graphite was purchased from Sigma-Aldrich.Carbon black was purchased from VWR. Lubricant PAO4 base oil was pur-chased from Exxon-Mobil. GOwasmade by amodified Hummersmethod (26)described previously (5, 27). Crumpled graphene balls were prepared bycapillary compression in rapidly evaporating aerosol droplets of GO sheets asreported previously (6). An ultrasonic atomizer (1.7 MHz, UN-511 AlfesaPharm Co.) was used to generate aerosol droplets of aqueous GO solution ata concentration of 1.5 mg/mL. Nitrogen flow was used to carry thosedroplets through a 400 °C tube furnace. Particles were collected at the endof the tube furnace using a Millipore Teflon filter with 200-nm pore size (6).Those partially reduced crumpled GO particles were further reduced at700 °C in argon for 1 h. r-GO was synthesized by hydrazine reduction ofGO in water and collected by filtration based on a previous report (28).

Tribology Tests. A pin-on-disk tribometer was used to study the tribologicalproperties. The steel disks for friction tests were machined from an E52100steel bar, and the disk surfaces were machine-polished to a mirror finish withsurface roughness Ra of around 5 nm measured by an interferometer. Thesteel balls, 3/8“ in diameter and made of M50 steel, were purchased fromMcMaster-Carr and used as received. Lubricant additives (graphite, carbonblack, and crumpled graphene balls) were added to the PAO4 base oil(density = 0.82 g/mL) and sonicated for 30 min in a water-bath ultrasoniccleaner UC-32D, 125 W. Due to its poor dispersibility, the filtered r-GO wastip-sonicated (150 W) for 10 min before sonicating in a water bath for

20 min. Before testing, the polished 52100 steel disks and steel ball were son-icated in acetone for 5 min to remove any possible residual contaminants.The metal disk was then fixed tightly in the holder of the tribotester, andplastic pipettes were used to transfer 3 mL of freshly mixed lubricant solu-tion onto the disk. The tests were conducted at a linear speed of 10 mm/s, aconstant vertical force of 10 N (maximum Hertzian contact pressure ∼1 GPa),and ambient temperature and humidity. The experimental duration was2,000 s and 4,000 s, respectively, for the 0.01 wt % and 0.1 wt % concen-tration of each carbon-based additive. Each sample was tested at least twiceunder identical conditions.

Characterization of Wear Tracks. Before each SEM observation, the metal diskwas cleaned in hexane for 3 min to remove the residual lubricant oil, and wasthen air-dried. SEM images were recorded using an LEO 1525 microscope.Before optical profilometry, the steel disk was further sonicated in acetone tocompletely remove all of the debris and lubricant materials. A Zygo NewView7300 optical surface profiler was used to identify and analyze the 3D to-pography of the wear track. The wear volume was defined as the amount ofmetal removed from a single track in the course of an experiment, and wasestimated by numerically integrating the surface height (from optical pro-filometry) over the area at eight different points along the track. Wearcoefficient is given by the following equation:

Wear  coefficentðKÞ=Wear  volume�m3

�× Surface  hardnessðPaÞ

Normal  loadðNÞ× Sliding  distanceðmÞ .

Vickers hardness measurements of steel disks were determined to be 575 ±10.4 kgf/mm2 (∼5.64 GPa) by a Struers Duramin microhardness tester. Themeasurements were repeated three times for each disk.

ACKNOWLEDGMENTS. We thank Drs. F. Guan, J. Luo, and K. Raidongia fortheir earlier exploration related to this work. The work was supported by theOffice of Naval Research (ONRN000141310556). J.H. thanks the John SimonGuggenheim Memorial Foundation for a Guggenheim Fellowship. H.D.J.was supported by the Basic Research Project of the Korea Institute of Geo-science and Mineral Resources, funded by the Ministry of Science, ICT andFuture Planning.

1. Bakunin VN, Suslov AY, Kuzmina GN, Parenago OP (2004) Synthesis and application of

inorganic nanoparticles as lubricant components - A review. J Nanopart Res 6(2-3):273–284.2. Wang X-B, Liu W-M (2013) Nanoparticle-based lubricant additives. Encyclopedia of

Tribology, eds Wang QJ, Chung Y-W (Springer, New York), pp 2369–2376.3. Hsu SM (2004) Nano-lubrication: Concept and design. Tribol Int 37(7):537–545.4. Bender JW (2013) Friction modifiers. Encyclopedia of Tribology, eds Wang QJ,

Chung Y-W (Springer, New York), pp 1355–1359.5. Luo J, Kim J, Huang J (2013) Material processing of chemically modified graphene:

Some challenges and solutions. Acc Chem Res 46(10):2225–2234.6. Luo J, et al. (2011) Compression and aggregation-resistant particles of crumpled soft

sheets. ACS Nano 5(11):8943–8949.

7. Luo J, Jang HD, Huang J (2013) Effect of sheet morphology on the scalability of

graphene-based ultracapacitors. ACS Nano 7(2):1464–1471.8. Narayan R, Kim S (2015) Surfactant mediated liquid phase exfoliation of graphene.

Nano Convergence 2(1):20.9. Berman D, Erdemir A, Sumant AV (2014) Graphene: A new emerging lubricant. Mater

Today 17(1):31–42.10. BollmannW, Spreadborough J (1960) Action of graphite as a lubricant.Nature 186(4718):29–30.11. Huang HD, Tu JP, Gan LP, Li CZ (2006) An investigation on tribological properties of

graphite nanosheets as oil additive. Wear 261(2):140–144.12. Lee CG, et al. (2009) A study on the tribological characteristics of graphite nano lu-

bricants. Int J Precis Eng Manuf 10(1):85–90.

Fig. 5. Comparison of lubrication between crumpled graphene balls in PAO4 base oil and commercial lubricant 5W30. At 0.1 wt % of loading level, the PAO4base oil modified by crumpled graphene balls outperforms the fully formulated lubricant 5W30 (additives up to 10 wt %) as shown in the comparison of (A)coefficient of friction, (B) wear coefficient, and (C and D) 3D profile images of the wear tracks.

1532 | www.pnas.org/cgi/doi/10.1073/pnas.1520994113 Dou et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

13. Lin JS, Wang LW, Chen GH (2011) Modification of graphene platelets and their tri-bological properties as a lubricant additive. Tribol Lett 41(1):209–215.

14. Mungse HP, Khatri OP (2014) Chemically functionalized reduced graphene oxide as a novelmaterial for reduction of friction and wear. J Phys Chem C 118(26):14394–14402.

15. Zhang W, et al. (2011) Tribological properties of oleic acid-modified graphene aslubricant oil additives. J Phys D Appl Phys 44(20):205303.

16. Baik S, et al. (2009) Frictional performances of activated carbon and carbon blacks aslubricant additives. Tribol Trans 52(1):133–137.

17. Hardy WB (1922) Boundary lubrication - The paraffin series. Proc R Soc Lond A 100(707):550–574.

18. Bhushan B (2012) Boundary lubrication. Encyclopedia of Nanotechnology, edBhushan B (Springer, Dordrecht, The Netherlands), pp 354–362.

19. Udofia IJ, Jin ZM (2003) Elastohydrodynamic lubrication analysis of metal-on-metalhip-resurfacing prostheses. J Biomech 36(4):537–544.

20. Hess DP, Soom A (1991) Normal vibrations and friction under harmonic loads. 1.Hertzian contacts. J Tribol 113(1):80–86.

21. Willis JR (1966) Hertzian contact of anisotropic bodies. J Mech Phys Solids 14(3):163–176.

22. Silbert LE (2010) Jamming of frictional spheres and random loose packing. SoftMatter 6(13):2918–2924.

23. Ronen A, Malkin S (1981) Wear mechanisms of statically loaded hydrodynamicbearings by contaminant abrasive particles. Wear 68(3):371–389.

24. Dwyerjoyce RS, Sayles RS, Ioannides E (1994) An investigation into the mechanisms ofclosed 3-body abrasive wear. Wear 175(1-2):133–142.

25. Prabhakaran A, Jagga CR (1999) Condition monitoring of steam turbine-generatorthrough contamination analysis of used lubricating oil. Tribol Int 32(3):145–152.

26. Hummers WS, Offeman RE (1958) Preparation of graphitic oxide. J Am Chem Soc80(6):1339.

27. Kim F, et al. (2010) Self-propagating domino-like reactions in oxidized graphite. AdvFunct Mater 20(17):2867–2873.

28. Li D, Müller MB, Gilje S, Kaner RB, Wallace GG (2008) Processable aqueous dispersionsof graphene nanosheets. Nat Nanotechnol 3(2):101–105.

Dou et al. PNAS | February 9, 2016 | vol. 113 | no. 6 | 1533

ENGINEE

RING

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020